What proportion of lung cancers can be operated by segmentectomy? A computed-tomography-based simulation

Abstract

OBJECTIVE

To estimate the probability that a lung cancer arising in a segment has safety anatomical margin for segmentectomy by using a computed-tomography (CT)-based simulation technique.

METHODS

We measured the volume of each segment by dividing a three-dimensional lung model into a segment model. We also measured the volume of particular portions of each segment that were located away from the intersegmental plane by a predefined distance according to a virtual tumor size of 1, 2, or 3 cm. The probability that a lung cancer arising in the segment has safety anatomical margin for segmentectomy (chance to accept segmentectomy) was expressed as the ratio of this particular portion to the entire segment.

RESULTS

There was significant variability in segment size (smallest, the right medial-basal; largest, the left apicoposterior segment). The chance to accept segmentectomy depended on the segment size and the virtual tumor size; however, irrespective of segment size, there was only a small chance to accept segmentectomy in the bilateral lateral-basal and left anterior segments. Overall, the chance to accept segmentectomy for virtual tumors of 1, 2, and 3 cm in diameter was 33%, 24%, and 18%, respectively. Bisegmentectomy provided 49% chance in resecting virtual tumors that were 2 cm in diameter.

CONCLUSION

The chance to accept segmentectomy differed greatly in individual segments; it was minimal if a segment was small or located between neighboring segments. Bisegmentectomy can increase the chance to accept segmentectomy. In addition to these results, our method is useful in identifying tumors having eligibility for segmentectomy.

Segmentectomy, Pulmonary resection, Lung cancer, Computer tomography, Surgical margin

INTRODUCTION

Segmentectomy can become a radical operative procedure for some patients with small (<2 cm in diameter) peripheral-type lung cancer [1–4]. Phase III trials comparing the long-term survival of patients receiving segmentectomy with lobectomy for peripheral-type lung cancer are currently being conducted in the United States and Japan [5]. In general, segmentectomy is indicated if a tumor is located >2 cm from the intersegmental plane and located in the outer one-third of the lung field on a chest computed tomography (CT) image [5]. However, it remains unclear whether these requirements are indeed applied because there is no definitive method to precisely measure the distance between the tumor and the intersegmental plane. In this article, we introduce a novel method to accurately measure the distance between the tumor and the intersegmental plane. We found that the likelihood to accept segmentectomy depends on the anatomical property of the segment itself. The aim of the present study was to clarify what part of the lung can predominantly be resected by segmentectomy by measuring the anatomical properties of each segment. We also examined whether a tumor needed to be located in the outer one-third of the lung field to preserve safe anatomical margins for segmentectomy.

PATIENTS AND METHODS

Patients

This study was approved by the Institutional Review Board of the Yamaguchi University School of Medicine. Ten adult patients who received chest CT scans for suspected lung cancer were subjected to regional lung volume measurements. Patients with evident pulmonary emphysema on chest CT, representing a low-attenuation area (attenuation value < −910 Hounsfield units (HU)) of more than 10% of the entire lung, were not included because segmentectomy may not have been preferable to lobectomy. In these patients, segmentectomy may have resulted in bronchoalveolar air leaks and lobectomy would likely result in an improvement of remaining lung function (volume reduction effect) [5]. Patient characteristics are shown in Table 1.

Table 1:

Patient characteristics (n = 10)

Variable	No. or mean ± SD
Age (years)	66.5 ± 13.5
Sex (M/F)	6/4
Height (cm)	164.4 ± 12.0
Smoking history (yes/no)	5/5
FVC (l)	3.64 ± 1.02
FEV1 (l)	2.68 ± 0.83
FEV1/FVC	73.6 ± 5.4
% FEV1 (%)	111.4 ± 13.1
Lung volume (l)	5.36 ± 1.15

FVC: forced vital capacity; FEV1: forced expiratory volume in 1 s; and %FEV1: % of predicted FEV1 by age, sex, and height

CT scans

CT scans were performed using a 64-detector-row CT scanner. With the patient in the supine position, we obtained 1-mm-thick, high-resolution CT images covering both lungs entirely in a 512 × 512 matrix during a deep inspiratory breath-hold. We used 1 mm collimation with a scan time of 1.0 s at 120–130 kVp and 220–230 mA. Transaxial CT images were reconstructed using the lung algorithm.

Threshold limits of −600 to −910 HU were applied to segment both lungs and to exclude soft tissue surrounding the lung and the large vessels, atelectasis, fibrosis, lung tumors, and low-attenuation areas within the lung [7–9]. Three-dimensional (3D) lung volume images representing lung contours, including lobar fissures, vasculatures, and volume, were created using imaging software (M900 QUADRA, Zio Soft K.K., Osaka, Japan). With this simulation software, the 3D lung model can be freely rotated on the computer screen, and any portion of the lung can be removed from the screen by outlining that area directly on the 3D lung model using a free-hand pen tool, simulating a real pulmonary lobectomy or segmentectomy [6]. In particular, the intersegmental plane is readily depicted by tracing the intersegmental pulmonary vein or segmental fissure and cutting the lung model in accordance with these anatomical structures. The distance between a tumor and the intersegmental plane can be measured by drawing an orthogonal line to the intersegmental plane via the tumor (Fig. 1, Supplementary Data). With this method, bilateral lungs of the 10 patients were divided into lobes. Bilateral upper and lower lobes were further divided into single segments. The volume of each segment was recorded and expressed as the percentage of standard values. The standard volume of a segment was defined as 1/18 of the total lung volume. To predict the chance to accept segmentectomy in individual segments, we measured specific portions of each segment using a predefined length from the intersegmental plane according to the virtual tumor size (Fig. 2). The chance to accept segmentectomy is expressed as the ratio of the volume of each specific portion to that of the entire segment. Because the chance to accept segmentectomy depends on the diameter of the tumor, we predicted the chance according to whether the virtual tumor measured 1, 2, or 3 cm in diameter.

Figure 1:

Computed tomography (CT) image showing a small peripheral lung adenocarcinoma in the anterobasal segment of the left lower lobe adjacent to the superior segment (A). A 3-dimensional CT model of the affected lung lobe (B) and its subdivision (C). The intersegmental plane between the superior segment and the basal segment is outlined in yellow. This intersegmental plane contains the intersegmental vein (V6b and V6c). The anatomical margin is readily depicted by drawing an orthogonal line (in green) to the intersegmental plane via the tumor (C).

Figure 2:

To secure a surgical margin of 2 cm, the center of a tumor must be located away from the intersegmental plane (dashed line) by at least (2 + r) cm (r = radius of a tumor). Thus, the center of the tumor must be located within the gray area. The chance to accept segmentectomy of a virtual tumor of 2r in diameter in the segment can be calculated as the ratio of the gray area to the total area of the segment. The chance to accept segmentectomy depend on tumor diameter.

Statistical analysis

Values are expressed as mean ± standard deviation (SD). Linear regression analysis was used to test the relationship between the two continuous variables. P < 0.05 was considered significant.

RESULTS

Significant variability in segment size was observed. The right S7 was the smallest segment (mean, 45 ± 13% of standard) and the left S1 + 2 was the largest segment (146 ± 56% of standard) (Table 2). The predicted chance to accept segmentectomy for virtual tumors of 1, 2, and 3 cm in diameter in each segment is shown in Fig. 3. The predicted chance to accept segmentectomy for virtual tumors of 2 cm in diameter was higher (>40%) for right S3, right S6, left S1 + 2, and left S6; medium (15–40%) for right S1, right S2, right S8, right S10, left S8, and left S10; and lower (<15%) for right S7, right S9, left S3, and left S9. In the higher-probability group, even around 2 cm tumors located in the inner two-thirds of the lung field could be resected by segmentectomy, whereas in the lower-probability group, there was no chance of segmentectomy for tumors of 2 cm in diameter located in the inner two-thirds of the lung field (Fig. 4). There was a significant correlation between the segment volume (% of standard) and the chance to accept segmentectomy for virtual tumors 2 cm in diameter (R = 0.462, P < 0.0001, Fig. 5). The patient-based mean chance to accept segmentectomy for virtual tumors 2 cm in diameter ranged from 17.1% to 29.3%. This chance varied in relation to total lung volume (R = 0.855, P = 0.0016). Overall, the mean chance to accept segmentectomy differed in relation to virtual tumor size: 33 ± 15% for 1 cm virtual tumors, 24 ± 13% for 2 cm tumors, and 18 ± 12% for 3 cm tumors. Chance to accept bisegmentectomy for virtual tumors of 1, 2, and 3 cm in diameter are shown in Fig. 6. Overall, the mean chance to accept bisegmentectomy correlated with virtual tumor size (1 cm: 56 ± 8%; 2 cm: 49 ± 8%; and 3 cm: 42 ± 8%).

Figure 3:

Chance to accept segmentectomy for virtual tumors 1, 2, and 3 cm in diameter in individual segments. The chances of resection of a virtual tumor 2 cm in diameter was higher (>40%) for the right S3, right S6, left S1 + 2, and left S6; medium (15–40%) for the right S1, right S2, right S8, right S10, left S8, and left S10; and lower (<15%) for the right S7, right S9, left S3, and left S9. Overall, the mean chance to accept segmentectomy for virtual tumors of 1, 2, and 3 cm in diameter was 33 ± 15%, 24 ± 13%, and 18 ± 12%, respectively.

Figure 4:

Axial image of a chest CT in a 53-year-old man. The masked area (red color) shows a portion of the superior segment located more than 3 cm from the intersegmental plane. Thus, if the center of a tumor, 2 cm in diameter, is located within the masked area, the tumor can be resected by superior segmentectomy with an anatomical margin of 2 cm or more. Note that the tumor can be located within the inner two-thirds of the lung field.

Figure 5:

Correlation between segment size (% of standard) and chance to accept segmentectomy for virtual tumors 2 cm in diameter (Y = 5.414 + 0.181X, R = 0.462, P < 0.0001). Note that the chance to accept segmentectomy rises with increased segment size.

Figure 6:

Chance to accept bisegmentectomy for virtual tumors 1, 2, and 3 cm in diameter. The chance to accept bisegmentectomy for virtual tumors 2 cm in diameter was highest (64 ± 4%) for tumors in the left upper division. Overall, the mean chance to accept bisegmentectomy for virtual tumors 1, 2, and 3 cm in diameter was 56 ± 8%, 49 ± 8%, 42 ± 8%, respectively.

Table 2:

Volumetric contribution of individual segments

Segment		% of standard value (mean ± SD)
Right lung
S1	Apical	119 ± 42
S2	Posterior	92 ± 30
S3	Anterior	145 ± 50
S4 + 5		81 ± 21
S6	Superior	110 ± 27
S7	Medial-basal	45 ± 13
S8	Anterior-basal	94 ± 23
S9	Lateral-basal	100 ± 29
S10	Posterior-basal	117 ± 33
Left lung
S1 + 2	Apicoposterior	146 ± 56
S3	Anterior	142 ± 44
S4 + 5	Lingula	90 ± 18
S6	Superior	80 ± 27
S8	Anterior-basal	85 ± 37
S9	Posterior-basal	94 ± 24
S10	Lateral-basal	101 ± 45

DISCUSSION

We have introduced an easy and practical way to measure the distance between a tumor and the intersegmental plane when planning lung segmentectomy for peripheral lung cancer. Preoperative assessment of the anatomical margin (CT-defined margin) is important because an insufficient margin may result in positive cytology of the surgical stump or even local recurrence [10, 11]. CT-defined margins are also important because intraoperative assessment of surgical margins with finger palpation is difficult if a tumor is indistinct because of an intrapulmonary location, small size, or bronchioloalveolar-type pathology and if the operation is performed thoracoscopically.

To reduce the likelihood of an insufficient surgical margin, segmentectomy is indicated for relatively small tumors (<2 cm in diameter) that arise within the outer one-third of the lung field. Indeed, there is a very limited chance to accept segmentectomy for relatively large tumors (>2 cm in diameter), especially for those arising at the right S7, right S9, left S3, and left S9. By contrast, there is some chance to accept segmentectomy for tumors 3 cm in diameter if the tumor is located at the right S3, right S6, left S1 + 2, and left S6. Furthermore, there is a considerable chance to accept segmentectomy even for tumors 3 cm in diameter if the neighboring segment is also resected (bisegmentectomy). We should therefore reevaluate long-term survival after segmentectomy or bisegmentectomy in patients who have relatively large tumors (2–3 cm). We believe that local recurrence rates, which are known to be higher after segmentectomy than after lobectomy [12], can be reduced by careful selection of patients eligible for segmentectomy on the basis of anatomical margin.

With respect to tumor location, tumors arising in the right S7, right S9, left S3, and left S9 must be located within at least the outer one-third of the lung field for the tumor to be resected with a safe margin. By contrast, tumors arising in the right S3, right S6, left S1 + 2, and left S6 do not necessarily need to be located within the outer one-third of the lung field. In particular, tumors located within the inner two-thirds of the left upper division can frequently be resected with sufficient margins by left upper division segmentectomy. Thus, it may not be useful to diagnose whether the tumor is located within the outer one-third of the lung field if only the CT-defined margin is secured.

A possible limitation of the present study is that the discrepancy between the CT-defined and the actual intersegmental plane is not verified. During segmentectomy, the demarcation is identified by various methods based on the distribution of either dominant airways or dominant pulmonary arteries of the segment to be resected [13, 14]. We believe that regardless of what method is used to intraoperatively identify the intersegmental plane, our simulation will actually reflect the surgical procedure of segmentectomy because during surgery, intersegmental planes are divided predominantly along with intersegmental vein, which is preoperatively identified by 3D-CT angiography [15]. Nevertheless, we must be aware that not all segments have an intersegmental vein because of the anatomical variation of the pulmonary vein. Even in patients with some variation, we can identify anatomical demarcation by referring to both the 3D bronchial and vascular trees.

As mentioned previously, insufficient surgical margins have been identified as a risk factor for positive cytology of the surgical stump or local recurrence. It may be difficult, however, to accurately measure the length of the surgical margin after dissection of the intersegmental plane because the lung parenchyma has collapsed and has been manipulated by electric coagulation or stapling during dissection. Thus, we propose the routine measurement of CT-defined margins before segmentectomy. The significance of the CT-defined margin in the risk of positive cytology for the surgical stump should be clarified in the future.

According to our study, segmentectomy can be indicated for only one-fourth to one-third of patients with small lung cancers (<2 cm in diameter). To indicate segmentectomy in additional patients, one could perform extended segmentectomy or combined resection of neighboring subsegments. Whether additional resection of neighboring subsegments is sufficient for resection of lymphatic drainage, however, remains unclear and needs further validation.

We estimated the overall effectiveness of segmentectomy for virtual tumors of 1, 2, and 3 cm in diameter. This estimation, however, was based on the hypothesis that lung cancers distribute homogeneously to all lung fields, whereas they actually distribute heterogeneously. A previous report on resected primary lung cancer showed that the ratio of tumor distribution to the right upper lobe, the right superior segment, the right basal segment (S7–10), the left superior division segment (S1 + 2, S3), the left superior segment, and the left basal segments (S8–10) is approximately 10:2:3:7:1:2 [16]. If this likelihood of tumor location is taken into consideration, the overall chance of segmentectomy for virtual tumors of 1, 2, and 3 cm in diameter is 36.6%, 27.8%, and 21.0%, respectively.

In summary, the chance to accept segmentectomy differs greatly in individual segments; it is minimal if a segment is small or located between the neighboring segments. Bisegmentectomy can increase the chance to accept segmentectomy. Since segmentectomy can be applied to only a proportion of patients with small lung cancer, routine measurement of CT-based anatomical margin is recommended.

Conflict of interest: none declared.